Automation connector

The automation connector gives your AI agent full control over the video creation pipeline. Unlike the public connector, which stops at a prepared project, the automation endpoint can edit video in a headless browser, handle interactive decisions, and export the finished file, all without a human opening Rendley.

Endpoint

https://mcp.rendley.com/mcp/automation

This endpoint requests the elevated rendley:mcp:full OAuth scope. A token granted only rendley:mcp:prepare receives 403 insufficient scope. API keys have full access.

Available tools

The automation connector includes every tool from the public connector plus:

Tool	Description
`create_workspace`	Create a new workspace
`create_project`	Create a video project
`duplicate_project`	Copy an existing project
`get_account`	Check plan details, credit balance, and storage usage
`edit_video`	Start a headless editor session that interprets your instructions
`check_edit`	Poll an edit job’s status (long-polls up to 20 seconds)
`respond_edit`	Respond to an interactive pause (approve, reject, or answer)
`cancel_edit`	Cancel a running or paused edit
`export_project`	Start a background render (codec, resolution, quality)
`check_export`	Poll an export job and get the download URL when done
`get_motion_clip_guide`	Retrieve the motion-clip JavaScript authoring guide
`get_brandkit`	Read brand voice, colors, fonts, text styles, and assets
`add_brand_assets`	Upload logos, fonts, or media to the brand kit
`add_brand_colors`	Add named colors to the brand kit
`set_brand_voice`	Update brand voice (tone, audience, energy, notes)
`set_brand_text_style`	Set typography for a text role (title, subtitle, heading, body, label)

Tools shared with the public connector: list_workspaces, list_projects, get_project, delete_project, add_files, finish_upload.

Typical workflow

A full automated video creation follows this pattern:

Create a project with create_project.
Upload media with add_files and finish_upload (if the video needs source files).
Start editing with edit_video. Pass the project ID and your instructions. This returns a job_id.
Poll the edit with check_edit. The edit may pause for interactive decisions.
Handle pauses with respond_edit when the status is waiting_input. The editor pauses for plan review (approve/reject the editing plan), paid action approval, clarification questions, or motion-clip authoring.
Export with export_project once editing is complete. Returns an export_id.
Poll the export with check_export until done. The response includes a download URL.

Interactive pauses

The headless editor runs in interactive mode. It may pause at several points:

Plan review. Before starting edits, the agent presents its plan. Respond with "approve" or "reject" (with feedback).
Paid action approval. Actions that cost credits (AI generation, stock media) pause for confirmation.
Clarification. The agent needs more information to proceed. Respond with free text.
Motion-clip authoring. The agent can write JavaScript motion clips. If authoring is needed, it pauses with the clip context.

Use respond_edit with the job_id and your response to continue.

Connecting

Configure the automation endpoint the same way as the public connector, but use the /mcp/automation URL. See the setup guides for your client:

Claude. Use https://mcp.rendley.com/mcp/automation as the endpoint URL.
ChatGPT / Codex. Use the automation URL in your connector or config.
Other clients. Replace the endpoint URL in the JSON config block.

Keep the two connector configurations separate. Do not submit the automation endpoint to public connector directories.

Paid plan requirement

Editing and export tools require a paid Rendley plan. Free-plan users see the tools listed but every call returns an upgrade prompt. The get_account tool is always available regardless of plan.

The same API, database, object storage, Redis, and Cloudflare browser worker are reused; there is no separate service to deploy.